Biochem. J. (1997) 327, 847–851 (Printed in Great Britain) 847

RNA minihelices as model substrates for ATP/CTP:tRNA Zengji LI1, Yue SUN2 and David L. THURLOW3 Department of Chemistry, Clark University, 950 Main Street, Worcester, MA 01610, U.S.A.

Twenty-one RNA minihelices, resembling the coaxially stacked loop affected the interaction. In general, RNAs with purines in acceptor- \T-stems and T-loop found along the top of a tRNA’s the loop were better substrates than those with pyrimidines, three-dimensional structure, were synthesized and used as sub- although no single base identity absolutely determined the strates for ATP\CTP:tRNA from effectiveness of the RNA as substrate. RNAs lacking bases near Escherichia coli and Saccharomyces cereŠisiae. The sequence of the 5h-end were good substrates for the E. coli , but were in the loop varied at positions corresponding to poor substrates for that from yeast. The apparent Km values for residues 56, 57 and 58 in the T-loop of a tRNA. All minihelices selected minihelices were 2–3 times that for natural tRNA, and were substrates for both , and the identity of bases in the values for apparent Vmax were lowered 5–10-fold.

INTRODUCTION ments represent an initial step in a long-range plan to design RNA molecules that can act as potent competitive inhibitors of The enzyme ATP\CTP:tRNA nucleotidyltransferase interacts tRNA nucleotidyltransferase, which is an essential enzyme for specifically with tRNAs to restore all or part of the 3h-terminal rapidly growing eukaryotic (e.g. tumor) cells. sequence CCA [1]. The enzyme does not discriminate between tRNAs, and yet, in general, does not effectively recognize other types of RNA such as synthetic homopolymers or rRNA [1]. The EXPERIMENTAL importance of the 3h-terminal base(s) in the recognition process Cloning degenerate oligonucleotides has been demonstrated using tRNAs [2], model substrates [3], and even 5S rRNA [2]; although in the latter case the reaction Two synthetic, partially complementary oligonucleotides (Gen- with rabbit-liver enzyme exhibited a much lower rate than with osys Biotechnologies, The Woodlands, TX, U.S.A.) were de- natural tRNA as substrate. signed to code for an RNA minihelix resembling the acceptor- Phe The importance of other parts of the tRNA molecule in and T-stems of tRNA from E. coli and containing variable affecting the interaction has been demonstrated using fragments sequences in the loop: of tRNA [4,5], model substrates [6], chemically modified tRNAs SacII (sense) 5h [7–9], and in Šitro synthesized precursor tRNAs (p-tRNAs; [10]). CGGACTTGGTTNNNTTCCGAGTCCGCCCACCGCGGA Based on the chemical interference assays and mutant p-tRNAs 3h containing mutations in the T-loop, a two-site model for the interaction was proposed [7–10] in which the enzyme ensures (template) 3h GCCTGAACCAANNNAAGGCTCAGGCGGG specificity for tRNAs by extending across the top to the tRNA’s TGGCGCCTTCGA 5h three-dimensional structure, making contact in the vicinity of the where positions designated by N were fully degenerate. A T-loop, specifically near residues 56, 57 and 58, numbered recognition site for SacII was included adjacent to the 3h-end of according to Steinberg et al. [11]. the gene, so that the construct could be linearized before in Šitro To further test this hypothesis and to eliminate the com- . When hybridized, the oligonucleotides contained a plication of tertiary structural effects, we describe in this report blunt end at one terminus and a four-base 3h-overhang compatible the synthesis of RNA minihelices resembling the top of a tRNA’s with ends left following cleavage by HindIII at the other. tertiary structure, and the ability of such minihelices to act as The hybridized oligonucleotides were phosphorylated using substrates for nucleotidyltransferases from Escherichia coli and polynucleotide according to the supplier’s instructions Saccharomyces cereŠisiae. The minihelix RNAs vary in sequence (New England Biolabs, Beverly, MA, U.S.A.) and joined to the within the loop at positions analogous to bases 56, 57 and 58 in vector pTZ18U (US Biochemicals, Cleveland, OH, U.S.A.) using the T-loop of a tRNA. Accordingly, experiments were carried T4 DNA (New England Biolabs). The pTZ18U had out to determine specifically: (1) if RNA minihelices can serve as previously been cleaved with EcoRI, treated with nuclease S1 (to substrates for tRNA nucleotidyltransferase; (2) if the identity of remove single-stranded overhanging ends), extracted with phenol bases in the T-loop of a tRNA analogue can affect recognition, [phenol\chloroform\isoamyl alcohol (49.5:49.5:1, by vol.) buf- independently from effects on tertiary structure; and (3) if the fered with 10 mM Tris\HCl (pH 8.0)\1 mM EDTA; Interna- best minihelix substrates have kinetic parameters Km and Vmax tional Biotechnologies, New Haven, CT, U.S.A.], digested with that are comparable with those for intact tRNA. These experi- HindIII, and agarose-gel purified using NA45 DEAE membranes

Abbreviation used: p-tRNAs, precursor tRNAs. 1 Current address: Department of Molecular Genetics and Microbiology, University of Massachusetts Medical Center, 55 Lake Avenue North, Worcester, MA 01655, U.S.A. 2 Current address: Department of Immunology and Microbiology, Bowman Gray School of Medicine, Medical Center Blvd., Winston-Salem, NC 27157, U.S.A. 3 To whom correspondence should be addressed. 848 Z. Li and others

(Schleicher and Schuell, Keene, NH, U.S.A.) to recover the RESULTS DNA. All restriction enzymes were obtained from New England Collection of analogues Biolabs. Relevant portions of the resulting recombinant vector and The collection of minihelices generated by cloning degenerate resulting transcript are shown below, including the start-site for synthetic oligonucleotides is depicted in Figure 1. The helix transcription by T7 RNA (*), the bases derived from resembles the coaxially stacked acceptor- and T-stems of the synthetic oligonucleotides (underlined), and the cut site (6) tRNAPhe from E. coli and the variable bases correspond to on the template strand for the unique SacII site introduced by positions 56, 57 and 58 in the T-loop. Over 100 recombinant the synthetic oligonucleotides. plasmids were selected at random and analysed. Nineteen of the * possible 64 different complete genes coding for minihelices were Sense: TAATACGACTCACTATAGGGCGGACTTGGTTN identified by sequencing, transcribed and used as substrate for NNTTCCGAGTCCGCCCACCGCGGAAGCTT nucleotidyltransferase. In this collection, all four bases are present at each of the variable positions. In addition, Template: ATTATGCTGAGTGATATCCCGCCTGAACCA- two constructs (AAA* and CGG*) were isolated that lacked the ANNNAAGGCTCAGGCGGGTGGCGCCTTCGAA three 5h-bases (GGG). These last three bases were originally 6 (SacII) present in pTZ18U, adjacent to the site of cleavage by EcoRI; RNA transcript: GGGCGGACTTGGTTNNNTTCCGAGT- they may have been removed during treatment with nuclease S1. CCGCCCACC The resulting RNA minihelix lacked a 3h-terminal adenosine Polyacrylamide gel assay residue, corresponding to the A of the CCA sequence normally found at the 3h-terminus of a tRNA. All RNA minihelices were substrates for both enzymes. Labelling Recombinant plasmids were used to transform competent E. of 15 minihelices by the E. coli enzyme is shown in Figure 2; coli JM109 cells and were isolated from white colonies on plates containing ampicillin and X-Gal using mini-plasmid preparation kits from Promega (Madison, WI, U.S.A.). Plasmids were screened using EcoRI, SacII and HindIII. Those lacking the EcoRI site, and containing SacII and HindIII sites, were se- quenced using Sequenase 2.0 DNA sequencing kits from US Biochemicals.

In vitro transcription Following complete cleavage of the plasmid template with SacII and extraction with phenol, RNAs were transcribed in Šitro using the Ribomax system (Promega) and T7 RNA polymerase purified according to Zawadzki and Gross [12] from E. coli BL21\ pAR1219 [13]. RNA products were extracted with phenol, precipitated with ethanol, and purified by eluting from 20% (w\v) polyacrylamide gels or by spun-column chromatography using Sephadex G50 [14]. Concentrations of the minihelices were determined using absorption at 260 nm, assuming an absorbance Figure 1 Sequence of RNA minihelices synthesized in vitro and used as of 1.0 corresponds to 40 µg\ml of RNA, and were verified using substrates for nucleotidyltransferase 20% (w\v) denaturing polyacrylamide gels [100 mM Tris\ borate, 8 M urea; acrylamide\N,Nh-methylenebisacrylamide The RNA minihelices comprised the above sequence with different bases in the loop (NNN), (19:1, w\w)] stained with 0.05% (w\v) Stains-all. corresponding to positions 56, 57 and 58 in the T-loop of a tRNA. The 19 sets of variable bases are listed on the left. The remaining sequence resembles that of the acceptor- and T-stems of tRNAPhe from E. coli. Enzyme purification and assay E. coli and yeast nucleotidyltransferases were purified as de- scribed previously [10]. The final concentration of enzyme in the assay was approx. 0.025 units\ml (one unit of activity results in the attachment of 1 µmol of AMP to tRNA per h at 37 mC). The assay contained 50 mM Tris\glycine (pH 9.4), 10 mM MgCl#, $# 5 mM dithiothreitol, 500 µM[α- P]ATP (0.7 mCi\ml) and sub- strate, in a volume of 7.5–15 µl. Following incubation for 30–60 min, deionized formamide containing 0.025% (w\v) Bromophenol Blue and 0.025% (w\v) Xylene Cyanole was added; and the entire sample was loaded on to a 20% denaturing polyacrylamide gel. Incorporation of label remained linear with time over the duration of the assay. Autoradiograms were Figure 2 Autoradiogram of RNA minihelices labelled with 32P by nucleo- exposed at k70 mC for 1–3 days. These were used as guides to tidyltransferase from E. coli slice out radioactive bands, which were placed in a Packard Tri- Carb 1900 CA liquid-scintillation counter to determine radio- An unfractionated mixture of E. coli tRNAs from which the 3h-terminal A residue had been removed was used as reference (lane labelled ‘tRNA’). The minihelices are depicted by the activity. Densitometry scans of the autoradiograms were per- triplet of variable bases in the loop. Those indicated by an * lack the 5h-bases GGG. All RNAs formed using a Personal Densitometer SI (Molecular Dynamics, were present at a concentration of 33 µM. The arrow indicates the position of the labelled Sunnyvale, CA, U.S.A.). minihelix. Interaction of minihelix RNAs with nucleotidyltransferases 849

(a) (b)

Figure 3 Autoradiogram of RNA minihelices labelled with 32P by nucleo- tidyltransferase from yeast

An unfractionated mixture of E. coli tRNAs from which the 3h-terminal A residue had been removed was used as reference (lane labelled tRNA). The minihelices are depicted as described in the legend to Figure 2. All RNAs were present at a concentration of 3.3 µM. The arrow indicates the position of the labelled minihelix.

Table 1 Extents of labelling of 14 minihelix RNAs relative to that for the Figure 4 Autoradiogram of RNA minihelices labelled with 32P by nucleo- minihelix AAA tidyltransferase from yeast (a) or E. coli (b) in the presence of an equimolar Each minihelix RNA is depicted according to the variable triplet sequence in the loop; that amount of wild-type or mutant p-tRNAs marked with an * lacks the 5h three bases GGG. Labelled bands were sliced from the gel (see The sequence of the wild-type p-tRNA is G57A58. The single mutant (A57A58) was used with Figures 2 and 3) containing samples incubated at a substrate concentration of 3.3 µM and that from yeast (a); both single (A57A58) and double (A57U58) mutant p-tRNAs were used as counted. Densitometry scans were also recorded. The ratios of extents of labelling for competing substrates with the E. coli enzyme (b). Minihelices are depicted as described in the 14 minihelices relative to that for the minihelix AAA were calculated using both radioactivity and legend to Figure 2. All RNAs were present at equimolar concentrations of 17 µM(a)or7µM densitometry data; the average of the two values is presented. (b). The arrow indicates the position of the labelled minihelix; the labelled p-tRNAs (top band) and its two or three degradation products migrate well above the minihelices. Analogue E. coli Yeast

AAA 1.00 1.00 comparable with AAA; GCA and CUU were labelled to a lesser AAA* 1.51 0.09 AGA 1.39 0.79 extent; and UCU and UUG were poor substrates. GAG 1.06 0.45 AAU 0.71 0.70 Competition gels AGU 1.00 0.54 AUA 0.34 0.38 The RNA minihelices were not detectably labelled in the presence AUU 0.66 0.49 of an equimolar amount of unfractionated tRNA (from which UAA 1.10 0.56 the 3h-terminal A residue had been removed), except at very low CAG 1.13 0.67 concentrations where the tRNA substrate may have been de- CAU 0.62 0.44 pleted before the end of the incubation. However, using nucleo- CUA 0.68 0.49 UCC 0.36 0.38 tidyltransferase from yeast, the minihelix AGA effectively com- GCU 0.92 0.50 peted with a mutant p-tRNA (p-A57), containing a G-to-A CGG* 0.75 0.26 substitution at position 57 (Figure 4a). In contrast, labelling of the minihelix UCC by the yeast enzyme in the presence of p-A57 was barely detectable (Figure 4a). Using the E. coli enzyme, the minihelices AAA and AAA* (lacking three bases on the 5h-end) were labelled poorly in the presence of wild-type p-tRNA, somewhat better in the presence of the single mutant p-A57, and labelling of the same set of 15 minihelices by the yeast enzyme is better still in the presence of a double mutant p-tRNA (p- shown in Figure 3. The labelled bands were sliced from the gel A57U58) (Figure 4b). and radioactivity was measured for each. In addition, den- sitometer scans of autoradiograms were performed. The minihelix Kinetic parameters AAA was a good substrate for both enzymes and was selected as a reference, so that the relative effectiveness of RNA minihelices To assess whether some of the best minihelix substrates (AAA, as substrates could be assessed. As a specific example, ratios of AAA* and AGA) were comparable with intact tRNA in terms of band intensities (using both radioactivity and densitometry data) kinetic parameters, an additional set of substrate titration at the lowest concentration of substrate (3.3 µM) for the other experiments was carried out on a single polyacrylamide gel using 14 RNAs relative to that for AAA were calculated and are E. coli nucleotidyltransferase. Apparent Michaelis–Menten con- presented in Table 1. These ratios reproducibly depended on the stants (Km) and Vmax values were calculated using Eadie–Hofstee identity of the variable triplet sequence in the loop over the entire plots of radioactivity recovered from gel slices. In addition, range of substrate concentrations (3.3–33.3 µM). In a separate densitometry data from the initial set of four gels in which 15 experiment, six additional minihelix RNAs were tested using analogues were screened for activity over a range of substrate only the yeast enzyme. The extents of labelling in this particular concentrations (see Figures 2 and 3) were analysed using initial experiment were not quantified; however, visual estimations of rates of incorporation into intact tRNA as an internal reference. band intensities suggest that GAU and UUA RNAs were The kinetic parameters for each selected minihelix were calculated 850 Z. Li and others

Table 2 Apparent Km and Vmax values for selected minihelices using loop of a tRNA to the recognition process [7–10]. To test the nucleotidyltransferase from E. coli hypothesis that nucleotides at positions analogous to 56, 57 and Values for kinetic parameters represent an average using both radioactivity and densitometry 58 in the T-loop of a tRNA directly affect recognition, we Phe data. Apparent Km and Vmax values for E. coli tRNA and in vitro transcribed p-tRNAs have constructed RNA minihelices with different bases at these been described previously [10] and are included for reference. positions. The identity of bases in the loop reproducibly affected the extent of labelling by both enzymes (Table 1). As a general −1 RNA Km (µM) Vmax (µmol:h ) rule, RNA substrates with all purines at the three variable positions were better substrates than those with all pyrimidines. AAA 13 0.023 However, there was no unambiguous correlation between the AAA* 15 0.031 identity of bases at any one site with the level of activity. AGA 8 0.015 It is interesting to note that the best substrates (e.g. AAA, Phe tRNA 5 0.14 AGA, GAG, UAA, CAG) contained a purine at position ‘57’ p-G57A58 8 0.2 p-A57A58 10 0.2 and the worst substrates (CUA, UCC, AUU and AUA) con- p-A57U58 14 0.08 tained pyrimidines at this site. These data correlate well with the observations that substituting pyrimidines at position 57 in mutant p-tRNAs dramatically inhibited the interaction [10]. Despite the fact that the purine at position 57 is not base paired in the crystal structure of a tRNA [15], it is not clear whether base substitutions in p-tRNAs had disrupted tertiary structural using both densitometry and radioactivity data and the results elements that were essential for recognition. Results of the are presented in Table 2. For all three types of substrate (tRNA, present set of experiments suggest that the identity of bases p-tRNA and selected RNA minihelices), the apparent Km values within the T-loop can directly affect recognition by nucleotidyl- are similar; they differ at most by a factor of 3. Values of Vmax , because there can be no long-range tertiary inter- were reduced by a factor of 5–10, relative to that for intact tRNA actions within the RNA minihelix comparable with those in a (Table 2). tRNA. However, stacking or base pairing within the loop may also be DISCUSSION important. For example, purines would be more likely to form base pairs with other bases in the loop, which are all uridines Analogues are substrates (Figure 1). The nature of base pairing within the loop would All RNA minihelices were detectably labelled by either enzyme directly affect the pattern of hydrogen-bond donors\acceptors at substrate concentrations in the 3–33 µM range. The RNA and hydrophobic\hydrophilic surfaces available for interaction minihelices contained the sequence CpC at their 3h-end, and with the enzyme. Future molecular modelling studies on possible thereby resembled a tRNA lacking its 3h-terminal adenosine. loop conformations may help to characterize such patterns. They also conformed to the optimal 3h-terminal recognition The identity of the base at the site analogous to position 56 in sequence determined using tRNA and model substrates [2,3,6]. a tRNA did not appear to be decisive in determining whether the That is, small molecules such as CpC and cytidine have been RNA could act as substrate for either enzyme. That is, at least shown to be substrates for the enzyme from rabbit liver. However, one minihelix that contained each of the four possible nucleotides the apparent Km for CpC was 5 mM (compared with 4 µM for at the site analogous to position 56 was nearly comparable in tRNA) and the Vmax value was reduced nearly 20-fold [3]. Some activity (i.e. 75% or greater) with the AAA minihelix (e.g. AGA, of the minihelix RNAs have apparent Km values much closer to GAG, UAA and CAG; Table 1). that for tRNA (Table 2). It is therefore apparent that structural From crystal data, it is known that the universally conserved aspects of the RNA minihelices enhance substrate binding, cytidine at position 56 is involved in a tertiary base pair with G19 relative to simple mono- or di-nucleosides. [15]. Previous results of chemical interference assays indicated None of the RNA minihelices were able to effectively inhibit the importance of chemically intact residues at positions 56 [8] labelling of natural tRNAs; the best substrates had apparent Km and 19 [7]. These data are consistent with the results reported values roughly 2–3 times that for tRNAPhe (Table 2). Accordingly, here if formation of the tertiary C56–G19 base pair in a tRNA additional structural aspects of a tRNA, not present in the RNA is one of several structural features that contribute to interaction minihelices, may be required for optimal binding of substrate with nucleotidyltransferase. Such tertiary structure would not be and\or full activation of catalytic activity. We note that the E. found in the RNA minihelices; hence the identity of the base at coli enzyme requires a chemically intact G at position 19 in the the site analogous to position 56 should not greatly affect D-loop for full activity [7]. Perhaps additional interactions take recognition. In addition, the RNA minihelices would not be place at the corner of the tRNA that involve tertiary base-paired expected to compete effectively with the natural tRNA substrate, residues such as G19–C56. because they lack the tertiary structural element that contributes However, some minihelices were able to effectively compete at to recognition. Even the best minihelix substrates did not inhibit equimolar concentrations with mutant p-tRNAs, whose apparent labelling of a wild-type p-tRNA (Figure 4). Km values were somewhat higher than those for wild-type p- tRNAs (Figure 4). Thus the possibility still remains that an RNA Missing bases at the 5 -end affect yeast but not E. coli enzyme minihelix with a slightly different structure (i.e. a different h sequence in the loop or smaller overall length) might serve as an Relatively large fragments of tRNA have been shown to re- effective inhibitor of tRNA nucleotidyltransferases. anneal to form substrate for the yeast enzyme, provided that each fragment contained either the 5h- or the 3h-terminus of the tRNA [4]. Synthetic oligonucleotides and small fragments of Identity of nucleotides in the loop affects recognition tRNA have also been shown to be substrates for the yeast Several experiments using chemical modification and mutant p- enzyme, provided that there was some base pairing of the tRNAs have suggested the importance of nucleotides in the T- oligonucleotides that resembled the terminus of the acceptor Interaction of minihelix RNAs with nucleotidyltransferases 851 stem of a tRNA [5]. Accordingly, the importance of secondary ability of the minihelices to act as substrates. The lack of base- structure to substrate recognition was proposed. Our results paired nucleotides near the site of catalysis inhibits interaction support the notion that base pairing near the site of catalysis is with the yeast enzyme, but not with that from E. coli. important in the case of the enzyme from yeast, because constructs lacking the 5h three bases (AAA* and CCG*), which This work was supported by a grant to D.L.T. from the National Institutes of Health, were unable to form base pairs near the site of addition of AMP, F1R15GM44309-01A1. were very poor substrates for the yeast enzyme. In contrast, the same substrates were labelled very effectively by the enzyme from E. coli, suggesting that the bacterial enzyme REFERENCES is much less stringent in its requirement for base pairing near the 1 Deutscher, M. P. (1982) The Enzymes 15, 183–215 site of catalysis. These data are in accord with previous damage- 2 Deutscher, M. P. (1973) J. Biol. Chem. 248, 3116–3121 selection experiments in which chemically modified bases near 3 Masiakowski, P. and Deutscher, M. P. (1977) FEBS Lett. 77, 261–264 the 3h-end (positions 71–73) affected recognition of substrate by 4 Overath, H., Fittler, F., Harbers, K., Thiebe, R. and Zachau, H. G. (1970) FEBS Lett. the yeast enzyme far more frequently than when the E. coli 11, 289–294 5 Wang, G.-H., McLaughlin, L. W., Sternbach, H. and Cramer, F. (1984) Nucleic Acids enzyme was used [7]. They also correlate well with the observation Met Res. 12, 6909–6923 that the E. coli initiator tRNA is unique in that it does not 6 Masiakowski, P. and Deutscher, M. P. (1979) J. Biol. Chem. 254, 2585–2587 contain base-paired residues at the end of the acceptor stem (i.e. 7 Spacciapoli, P., Doviken, L., Mulero, J. J. and Thurlow, D. L. (1989) J. Biol. Chem. 1–72). Thus the E. coli enzyme must be able to tolerate the lack 264, 3799–3805 of base pairs near the site of catalysis if it is to repair the initiator 8 Hegg, L. A. and Thurlow, D. L. (1990) Nucleic Acids Res. 18, 5975–5979 tRNAMet. 9 Spacciapoli, P. and Thurlow, D. L. (1990) J. Mol. Recognit. 3, 149–155 10 Li, Z., Gillis, K. A., Hegg, L. A., Zhang, J. and Thurlow, D. L. (1996) Biochem. J. 314, 49–53 CONCLUSIONS 11 Steinberg, S., Misch, A. and Sprinzl, M. (1993) Nucleic Acids Res. 21, 3011–3015 RNA minihelices resembling the top of a tRNA’s three-dimen- 12 Zawadzki, V. and Gross, H. F. (1991) Nucleic Acids Res. 19, 1948 sional structure are effective substrates for tRNA nucleotidyl- 13 Davanloo, P., Rosenberg, A. H., Dunn, J. J. and Studier, F. W. (1984) Proc. Natl. Acad. Sci. U.S.A. 81, 2035–2039 E coli K from . and yeast; their apparent m values are 14 Sambrook, J., Fritsch, E. F. and Maniatis, T. (1989) Molecular Cloning: A Laboratory only 2–3 times that of a natural tRNA and their Vmax values are Manual, 2nd edn., Cold Spring Harbor Laboratory, Cold Spring Harbor, NY reduced 5—10-fold. Variations in sequence within the loop, 15 Holbrook, S. R., Sussman, J. L., Warrant, R. W. and Kim, S.-H. (1978) J. Mol. Biol. corresponding to positions 56, 57 and 58 of a tRNA, affect the 123, 631–660

Received 8 April 1997/4 July 1997; accepted 9 July 1997